line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Alvis::NLPPlatform::Canonical; |
2
|
3
|
|
|
3
|
|
13
|
use strict; |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
145
|
|
3
|
3
|
|
|
3
|
|
13
|
use warnings; |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
1716
|
|
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
our $VERSION=$Alvis::NLPPlatform::VERSION; |
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub CleanUp |
9
|
|
|
|
|
|
|
{ |
10
|
|
|
|
|
|
|
|
11
|
0
|
|
|
0
|
1
|
|
my ($canonical, $preserveWhiteSpace) = @_; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
# my $tmp_str; |
14
|
|
|
|
|
|
|
|
15
|
0
|
0
|
|
|
|
|
if (!$preserveWhiteSpace) { |
16
|
0
|
|
|
|
|
|
warn "\nRemoving White Spaces\n"; |
17
|
|
|
|
|
|
|
|
18
|
0
|
|
|
|
|
|
$canonical =~ s/^[\s\t]*(<[^>]+>)[\s\t\n]*(\n)/$1$2/go; |
19
|
0
|
|
|
|
|
|
$canonical =~ s/(\n)[\s\t\n]*(<[^>]+>)[\s\t]*(\n)/$1$2$3/go; |
20
|
0
|
|
|
|
|
|
$canonical =~ s/(\n)[\s\t\n]*(<[^>]+>)[\s\t]*/$1$2/go; |
21
|
|
|
|
|
|
|
} |
22
|
0
|
|
|
|
|
|
$canonical =~ s/ |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
|
26
|
0
|
|
|
|
|
|
$canonical =~ s/<\/?list>/\n/go; |
27
|
0
|
|
|
|
|
|
$canonical =~ s/<\/?item>/\n/go; |
28
|
0
|
|
|
|
|
|
$canonical =~ s/<\/?canonicalDocument>/\n/go; |
29
|
0
|
|
|
|
|
|
$canonical =~ s/<\/?ulink[^>]*>//go; |
30
|
0
|
0
|
|
|
|
|
if (!$preserveWhiteSpace) { |
31
|
0
|
|
|
|
|
|
$canonical =~ s/\n+/\n/go; |
32
|
0
|
|
|
|
|
|
$canonical =~ s/^\n//go; |
33
|
0
|
|
|
|
|
|
$canonical =~ s/^\n$//go; |
34
|
|
|
|
|
|
|
} |
35
|
|
|
|
|
|
|
|
36
|
0
|
|
|
|
|
|
my $pos = 0; |
37
|
0
|
|
|
|
|
|
my $str = $canonical; |
38
|
0
|
|
|
|
|
|
my $pos_section = -1; |
39
|
0
|
|
|
|
|
|
my $pos_prec_section = 0; |
40
|
|
|
|
|
|
|
|
41
|
0
|
|
|
|
|
|
Alvis::NLPPlatform::XMLEntities::decode($str); |
42
|
0
|
|
|
|
|
|
$canonical = ""; |
43
|
|
|
|
|
|
|
|
44
|
0
|
|
|
|
|
|
while(($pos_section = index($str, "", $pos_prec_section)) > -1) { |
45
|
0
|
|
|
|
|
|
$canonical .= substr($str, $pos_prec_section, $pos_section - $pos_prec_section); |
46
|
0
|
|
|
|
|
|
chomp $canonical; |
47
|
0
|
0
|
|
|
|
|
if ($pos_section != $pos_prec_section) { |
48
|
0
|
|
|
|
|
|
$canonical .= "\n"; |
49
|
|
|
|
|
|
|
} |
50
|
0
|
0
|
|
|
|
|
if (!$preserveWhiteSpace) { |
51
|
0
|
|
|
|
|
|
$canonical =~ s/\n+/\n/go; |
52
|
0
|
|
|
|
|
|
$canonical =~ s/^\n//go; |
53
|
0
|
|
|
|
|
|
$canonical =~ s/^\n$//go; |
54
|
|
|
|
|
|
|
} |
55
|
|
|
|
|
|
|
|
56
|
0
|
|
|
|
|
|
push @Alvis::NLPPlatform::tab_end_sections_byaddr, (length($canonical) - 1); |
57
|
0
|
|
|
|
|
|
$pos_prec_section = $pos_section + 10; |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
|
61
|
0
|
|
|
|
|
|
return($canonical); |
62
|
|
|
|
|
|
|
} |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
1; |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
__END__ |